Dialog with Robots: Papers from the AAAI Fall Symposium (FS-10-05)
The Social Medium Is the Message
V. Groom
V. Srinivasan
C. Nass
R. Murphy
C. Bethel
Stanford University
vgroom@stanford.edu
Texas A&M University
vasant.s@gmail.com
Stanford University
nass@stanford.edu
Texas A&M University
murphy@cse.tamu.edu
Yale University
cbethel@gmail.com
Introduction
interaction.
Robots are being considered for applications where they
serve as proxies for humans interacting with another human, such as emergency response, hostage negotiation, and
healthcare. In these domains, the human (“dependent”) is
connected to multiple other humans (“controllers”) via the
robot proxy for long periods of time. The dependent may
want to interact with humans but also to engage the robot
as a medium to the World Wide Web. In the future, medical personnel may use the robot for victim assistance and
comfort (i.e., a “survivor buddy” such as seen in Fig. 1(a))
while the rescue team plans and monitors extrication. Other
applications include healthcare, where the robot is the link
between a patient and a medical provider for intermittent,
routine interactions, and hostage negotiation, where police
may use a bomb squad robot to talk with and build rapport
with the suspect while the SWAT team uses the robot’s sensors to build and maintain situation awareness.
Under funding from the National Science Foundation, we
are finishing the first year of investigating verbal and nonverbal communication strategies for robots who are serving
as proxies for multiple humans interact with the humans who
are dependent on them (Fig. 2). Our work posits that such
a robot would occupy a novel social medium position according to the Computers as Social Actors (CASA) model
(Nass, Steuer, and Tauber 1994) (Reeves and Nass 1996).
Given that teleoperated robots are treated socially, it is unlikely that a rescue robot would be treated as a pure medium
even if playing music or videos. Likewise, the limitations
of autonomy and the interactions of specialists with the dependent prevent the robot from being a true social actor.
Instead, social actor and pure medium are two extremes on
the agent identity spectrum, with a social medium occupying a middle position. A social medium would be perceived
as a loyal, helpful “go between” who is an advocate for the
dependent, rather than a device for accomplishing the goals
of multiple controllers (medical specialist, structural engineer, rescue operations official, etc.). To explore the social
medium identity, we have built a physical prototype of a
Survivor Buddy and are creating autonomous affective behaviors and a social medium toolkit to explore human-robot
Current Formative Studies
(a) Simulation of Survivor Buddy (b) Simulation of Survivor
in SARGE
Buddy in Microsoft Robotics
Developer Studio
Figure 1: Survivor Buddy in Simulation
Two formative studies have recently been conducted Stanford’s CHIMe lab using Texas A&M’s simulation of the Survivor Buddy physical prototype (Fig. 1(b)).The studies were
set in the disaster response domain, where participants simulated being victims.
Study 1: Social Role and Framing
The first study examined the effects of robot social role and
framing on participants’ attitudes and behaviors. The study
featured a 3(role: pure medium v. social medium v. social
actor) x 2(framing: unframed v. framed) design.The pure
medium channeled the controllers directly without demonstrating a social presence, the social medium channeled the
controllers while demonstrating a social presence, much like
a dispatcher, while the social actor communicated with the
controllers but did not channel them directly. We hypothesized that participants would prefer a robot that demonstrated a social identity over a robot that presented itself as
a pure medium, as the social robots would offer companionship and minimize fear.We also manipulated framing, or
setting expectations. We hypothesized that informing participants of the robot’s social role before the interaction would
c 2010, Association for the Advancement of Artificial
Copyright Intelligence (www.aaai.org). All rights reserved.
139
Onboard
Autonomous
Affective
Behaviors
Role Identity
Mapping &
Collision
Resolution
Natural
Synthetic
Single
Synthetic
Dependent
(victim)
Survivor Buddy attachment
maintains consistent social
identity despite multiple
controllers
half they were perpendicular. The order of orientation was
counterbalanced, allowing us to determine if responses to
the independent variables depend on whether the robot approaches from the dependent’s side or from behind. The
simulated robot initiated the interaction by approaching the
participant and starting to talk about the disaster. In each
case, the robot had the same actions but different rotational
speeds of the joints (i.e., moving head faster and farther),
with greater expressiveness manifest with quicker, larger
movements. In order to determine cognitive load, tests of
creativity and memory were applied. To measure creativity,
participants completed an alternate uses task during the interaction. For example, the robot might ask ”The first object
is shoe. A shoe is usually used to put on your feet to help
walking. What else can a shoe be used for?” Higher number
of responses indicated greater creativity and lower cognitive
load. A memory task was also administered, where participants attempted to remember as many items from a list as
possible and restate them after forty-five seconds. In addition to these behavioral measures, self-report measures of
attitudinal responses to the robot were assessed with a questionnaire.
Both studies have been completed and the data is currently
being analyzed. Results from these studies will reveal the
effects of the following variables on dependents’ attitudes
and behaviors: social role, framing, expressiveness, and personal distance. These results will expand our understanding
of the effects of robot social behavior on dependents and
enable designers to create robots in the optimal social role,
demonstrating social behaviors that improve the experiences
of dependents. These studies were the first in a several year
program of study. Results will be incorporated into future
designs of Survivor Buddy. In addition, more studies featuring physical robots in a realistic survival settings will be
conducted to improve the ecological validity, allow for the
examination of the physiological impact in a realistically
stressful situation, and allow for replication of these results.
Social Medium Toolkit
Speech
Translator
Independent
Controllers with
Protocol Coach
Interfaces
Control of robot base arbitrated
by policies, priorities, etc. for
different roles
Figure 2: Overview of the Survivor Buddy testbed.
generate greater trust.
The study was conducted entirely online with eighty-four
participants. Student participants watched a video about a
disaster to set the scene, then, on the computer screen, an
animation of a rescue robot with the Survivor Buddy head
approached and initiated the interaction. Participants viewed
the robot from a trapped dependent’s perspective, low on
the ground in a collapsed room. Role was manipulated via
slight, but explicit, verbal differences indicating the social
identity. For example, the pure medium robot introduced itself by saying ”I am the controller of the robot that is here
to help you. I have information for you.” The social medium
said, ”I am a robot that is here to help you. I will be presenting information to you from my controller” and the social
actor said, ”I am a robot that is here to help you. I have
information for you.” Throughout the task, the robot recommended certain media clips, noting that typically different
media clips were recommended. This allowed us to record a
behavioral measure of compliance by summing the number
of times participants selected the robot’s recommendations
over the typical recommendations. A questionnaire was administered, assessing attitudinal responses of trust and fear.
Acknowledgements
This work was supported in part by NSF Grant IIS-0905485
“The Social Medium is the Message” and by a HRI grant
from Microsoft External Research.
Study 2: Expressiveness and Personal Distance
The second study extended the work by (Bethel, Bringes,
and Murphy 2009) that explored the effects of proxemics on
dependents. This study varied the level of expressiveness
of the robot and the distance between participants and the
robot in a 2(expressiveness: high vs. medium vs. low) x
2(distance: close vs. far) study design. We hypothesized
that low expressiveness would be preferred when the robot
is close, and high expressiveness when the robot is far, balancing the proxemics with the tendency to like highly expressive agents. Eighty-four participants watched a video of
a disaster situation to set the scene. They were then placed
on the floor in a dark room at the close or far distance from
a wall onto which a simulation of the robot was projected.
The close distance was two feet and the far distance was
six feet, representing personal and social proxemic zones
(Bethel and Murphy 2008). For half the interaction, participants were positioned parallel to the wall and for the other
References
Bethel, C., and Murphy, R. 2008. Survey of non-facial/nonverbal affective expressions for appearance-constrained
robots. Systems, Man, and Cybernetics, Part C: Applications
and Reviews, IEEE Transactions on 38(1):83–92. 10946977.
Bethel, C.; Bringes, C.; and Murphy, R. 2009. Nonfacial and non-verbal affective expression in appearanceconstrained robots for use in victim management (video and
abstract).
Nass, C.; Steuer, J.; and Tauber, E. 1994. Computers as
social actors. CHI 72–78.
Reeves, B., and Nass, C. 1996. The media equation: How
people treat computers, television, and new media like real
people and places.
140